Top-Down Corporate AI Rollouts Stall at 20% Adoption, McCoy Finds

Panel 5 - 4: From Fear to Fluency: AI Literacy Interventions as Workforce Readiness — Ai4

Ai4

Mandating artificial intelligence without addressing job security fears or providing guided prompt practice triggers silent resistance, capping workforce adoption at 20%.

AI-assisted summary of Panel 5 - 4: From Fear to Fluency: AI Literacy Interventions as Workforce Readiness — original source. Check the original for full context.

Source published:

Detailed notes

## Background Published on August 14, 2026, by the Ai4 channel on YouTube, this presentation features Kari McCoy, a doctoral candidate and career counselor at CSU Northridge. McCoy outlines findings from a pilot study showing how an AI industry panel and hands-on prompt engineering workshop transitioned continuation high school students from fear and misinformation to practical confidence, offering a scalable model for organizational AI literacy and workforce readiness.

## TL;DR McCoy addresses the growing divide between corporate AI optimism and widespread user apprehension, noting that unguided top-down rollouts often stall at roughly 20% adoption due to fear of job loss, ethical concerns, and frustrating early hallucinations. Drawing from her pilot study with historically underserved continuation high school students, she illustrates how demystifying AI via an expert panel and guided prompt engineering shifted 60% of students away from apocalyptic dread and into active, peer-led problem solving. Extending these insights to the workplace, she emphasizes that organizations must pair clear, compliant usage policies—preventing risks like HIPAA breaches and legal sanctions—with structured, low-stakes practice and peer champions. Ultimately, McCoy urges leaders to ensure their human support systems and psychological readiness initiatives evolve as rapidly as their technological implementations.

## Atmosphere & style McCoy strikes an empathetic, grounded, and practical tone, merging the compassionate insight of a seasoned counselor with the strategic clarity of an organizational advisor. The narrative charts an inspiring arc, progressing from descriptions of alienation, dread, and ethical confusion toward scenes of collaborative peer coaching and personal agency. Listening to the talk feels refreshingly honest and humane, puncturing insular tech-bubble assumptions to address the authentic anxieties of workers and vulnerable learners confronting algorithmic change.

## Key takeaways - [00:04](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=4s) AI literacy has emerged as the number-one skill sought by employers, making equitable training essential so underserved populations are not excluded from the evolving workforce. - [02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) Workforce skepticism and resistance stem from legitimate ethical concerns—including environmental resource consumption, intellectual property theft, and fear of job automation. - [02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) Without explicit organizational policies and psychological safety, employees and students remain trapped in uncertainty about whether using AI will be penalized or rewarded. - [04:41](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=281s) A pilot study with continuation high school students showed that combining an industry panel with hands-on prompt workshops successfully moves individuals from sci-fi dread to practical fluency. - [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) Guided, iterative prompt engineering for practical tasks like resumes and mock interviews empowers learners to set personal ethical boundaries and coach their peers. - [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) Top-down software mandates without guided practice typically stall at around 20% adoption, as unassisted workers quickly abandon tools after encountering initial hallucinations or errors. - [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) Proactively engaging employees across all organizational levels allows leadership to demystify deployment roadmaps and reassure staff that AI is meant to augment, not eliminate, their roles. - [13:14](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=794s) Sustainable adoption relies on pairing hesitant learners with peer champions, curating short tutorials, and offering competitive incentives like departmental innovation showcases. - [14:45](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=885s) Because legal accountability rests on the employer rather than third-party AI vendors, institutions must implement clear acceptable-use policies to prevent data privacy breaches and compliance violations. - [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) Leaders must ensure that technological deployments do not outpace human support systems, moving beyond passive suggestion boxes to provide dedicated, protected practice time.

## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=4s) Bridging the AI Literacy Divide in Workforce Adoption

Kari McCoy introduces her talk by connecting her perspective directly with the preceding presentations, focusing on a critical challenge confronting organizations: how to meaningfully increase employee adoption of artificial intelligence tools. Drawing on her multifaceted background as a veterans and career counselor across two California community colleges, a former K-12 educator and high school counselor with a decade of experience, a non-profit president, and an AI strategy and policy consultant, McCoy approaches the adoption question through an equity-centered lens.

She points out that while those attending tech conferences inhabit an insulated bubble where it feels like artificial intelligence is omnipresent and universally embraced, the reality on the ground is drastically different. AI is undeniably here and growing in everyday dependency, but widespread familiarity cannot be taken for granted. Historically underserved populations—particularly students from lower socioeconomic backgrounds—are systematically being left behind without equal access or foundational understanding, raising urgent questions about how they will compete as employers rapidly shift expectations.

- 86% of employers expect to transform their businesses through AI by 2030. - AI literacy has emerged as the number-one skill employers seek in incoming workers. - The fundamental workplace equity question: who is entering the labor force AI-literate and equipped with these tools, and who is being locked out?

---

[02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) Apprehension, Ethical Resistance, and the Need for Psychological Safety

Before leadership can realistically ask how to get a workforce to adopt AI, McCoy argues they must deeply examine where skepticism, fear, and active resistance originate. Within demographics like Generation Alpha and among workers more broadly, there is intense blowback fueled by real concerns: worries that AI infrastructure is consuming and polluting scarce water resources, draining critical energy reserves, facilitating academic plagiarism, and stealing intellectual property and livelihoods from creative artists. Many harbor the conviction that no usage of generative technology can be ethical. If leaders ignore this resistance and fail to establish clear guidelines, workers simply disengage or hide their usage.

McCoy stresses that sustainable adoption rests on cultivating self-confidence, defining practical applications, and establishing psychological safety. Workers need explicit assurance from leadership that learning and applying these tools will not automate away their positions. Currently, confusion dominates: students and employees alike encounter conflicting rules, where one teacher or manager permits AI while another bans it entirely. Without a transparent organizational policy, workers are trapped in uncertainty—wondering whether disclosing their AI usage will get them fired as redundant, or praised for higher productivity.

- Sources of user apprehension include environmental strain (water and power consumption by data centers), copyright theft, artist displacement, and personal data privacy risks. - Confusion stems from inconsistent institutional guidelines across classrooms and departments. - Leadership must articulate direct guardrails to alleviate the fear that efficiency gains will result in job elimination.

---

[04:41](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=281s) The Continuation High School Intervention: Demystifying the Black Box

To study how to move individuals from avoidance to literacy, McCoy conducted a qualitative pilot study with continuation high school students—a credit-deficient population that has fallen behind in traditional schooling and largely comes from historically underserved backgrounds. At the outset, the students exhibited a generalized, science-fiction-inspired dread, viewing AI as an uncontrollable force destined to take over the world. Many conflated the entire concept of artificial intelligence strictly with ChatGPT, possessing no awareness that everyday technologies like GPS navigation, Apple's Siri, or even specialized consumer applications rely on AI architectures.

The intervention was structured into two targeted components: an industry expert panel followed by a hands-on prompt engineering workshop. The panel convened AI professionals who spoke candidly about what the technology actually is, debunked what it is not, mapped its technical limitations, and discussed ethical pitfalls alongside workforce opportunities. Rather than treating AI as an untouchable mystery, the panel provided an open forum where students could ask unvarnished questions about their fears regarding data harvesting, environmental impact, and the destruction of artistic careers.

```mermaid flowchart TD accTitle: AI Literacy Intervention Model accDescription: Two-stage intervention model moving continuation students from fear to fluency.

subgraph Baseline["Initial Baseline"] A["Generalized Dread & Distrust<br/>'AI is Sci-Fi / ChatGPT only'"] end

subgraph Interventions["Two-Step Intervention"] B["Phase 1: Industry Expert Panel<br/>• Demystifying tech & limits<br/>• Frank ethical discussions"] C["Phase 2: Prompt Engineering Workshop<br/>• Hands-on mock interviews<br/>• Resume optimization & iteration"] end

subgraph Outcomes["Literacy Outcomes"] D["Practical Agency & Confidence<br/>• Active experimentation<br/>• Self-imposed ethical boundaries"] end

A --> B B --> C C --> D ```

---

[07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) Hands-On Prompt Mastery and the Transition to Active Agency

The second phase moved from passive observation to concrete skill-building through a hands-on prompt engineering workshop focused on high-stakes, practical tasks: conducting mock job interviews, refining resumes, and locating entry-level opportunities and scholarships. McCoy noted that early negative experiences had previously caused many students to abandon AI altogether; when an early prompt to ChatGPT yielded an inaccurate hallucination, they concluded the technology was useless hype. The workshop unlocked prompt iteration, teaching students how to provide context, refine outputs, and instruct the model to "sound more like me."

McCoy highlights three key empirical findings from the qualitative evaluations: 1. **Shattering the Sci-Fi Myth**: Prior to the intervention, 60% of students expressed fear that AI was an apocalyptic force destined to take over the world. Guided practice demystified the tool, shifting their view from magical threat to useful instrument. 2. **From Avoidance to Active Inquiry**: Students transitioned from "putting my head in the sand" to generating practical workforce applications, asking if the tool could tailor resumes to specific job postings or uncover college funding. 3. **Empowerment Through Ethical Boundaries**: Rather than staying paralyzed or resentful over intellectual property and privacy issues, students took ownership of their data. They articulated personal ethical lines regarding what information they would never share and where automated assistance crossed into dishonesty.

By the end of the workshop, students who mastered techniques early leaned over to assist their peers, coaching them with tips like, "No, no, try this way." This spontaneous peer teaching signaled a complete shift from alienation to self-efficacy, showing how guided, contextualized practice enables participants to define responsible usage for themselves.

[10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) Naming the Fear and Demystifying Workplace AI

When leaders fail to acknowledge employee anxiety around artificial intelligence, top-down rollouts inevitably trigger silent resistance. Simply announcing a tool by declaring, "Hey, we got a new tool. You're going to use it. It's going to be great. Let's be energized about it," yields low adoption—roughly 20%. While a handful of early adopters will jump in to showcase improvements, the vast majority quietly push back. Without structured guidance, hesitant employees try the tool tentatively, ask ChatGPT a single poorly framed prompt, receive an unhelpful hallucination or error, and immediately quit, concluding, "I knew this wouldn't work."

```mermaid flowchart TD accTitle: AI Adoption Pathways subgraph TopDown["Top-Down Mandate (~20% Adoption)"] A["Announce tool without guidance"] --> B["Unassisted employee prompts ChatGPT"] B --> C["Poor initial response"] C --> D["Silent resistance & abandoned use"] end

subgraph Guided["Empathetic Implementation (High Buy-In)"] E["Gather intel across levels: interns to engineers"] --> F["Validate fears & clarify scope"] F --> G["Guided practice with peer champions"] G --> H["Camaraderie & sustained adoption"] end ```

To prevent this cycle, leadership must meet employees where they are and proactively surface their apprehensions. Checking in across the organizational hierarchy—from interns to mid-level engineers—and asking how they genuinely feel allows leadership to gather vital intelligence and demystify the actual roadmap. When workers worry that a new tool means being replaced tomorrow, leaders can clarify the true scope: AI is being introduced to handle tedious work, boost individual productivity, and make the organization more competitive as a team, building shared camaraderie rather than resentment.

- Top-down mandates without psychological readiness typically cap out around 20% adoption due to unguided, frustrating initial experiences. - Gathering candid input across roles prevents worst-case assumptions about immediate job displacement.

---

[13:14](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=794s) Guided Practice, Champions, and Peer-Led Innovation

Moving from understanding to confidence requires intentional, low-stakes practice rather than simply handing employees a platform and saying "have at it." Unguided workers do not know where to start. Effective workforce upskilling relies on pairing hesitant staff with experienced counterparts, curating short instructional videos, surveying teams to identify existing organic use cases, and leveraging peer champions to share functional workflows.

Organizations can also tap into competitive incentives to drive adoption. For instance, a counseling department dean could host a showcase where counselors compete to identify the AI application that best enhances student outreach, with prizes awarded for practical solutions. Supplying software engineers or administrative staff with dedicated exploratory time and relatable peer role models transforms intimidating technology into an engaging, collaborative experiment.

- Unguided access leaves staff overwhelmed; pairing learners with peer counterparts provides a clear starting point. - Internal innovation showcases and departmental competitions incentivize employees to experiment and surface high-value workflows.

---

[14:45](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=885s) Institutional Ethics, Legal Liability, and Policy Clarity

Ethical training and concrete institutional policies must accompany technical rollouts, as legal accountability ultimately rests on the organization itself rather than the third-party software. In regulated fields like mental health counseling, well-meaning but untrained workers risk severe compliance breaches. For example, an intern might feed confidential client details into ChatGPT to automatically generate progress notes, inadvertently triggering severe HIPAA violations.

```mermaid flowchart LR accTitle: AI Risk Management Flow subgraph BlindUse["Ungoverned Usage"] R1["Sensitive data input / unverified research"] --> R2["HIPAA violations & legal sanctions"] end subgraph PolicyGovernance["Institutional Governance"] P1["Clear, evolving AI acceptable-use policy"] --> P2["Training on hallucinations & data privacy"] end ```

The perils of unguided usage are equally visible in the legal sector, where an attorney submitted court briefs citing ChatGPT-generated precedents, only to face heavy financial sanctions when the cited case law proved entirely fabricated. Because everyday workers rarely understand baseline limitations regarding data privacy or model hallucinations, leaders must establish an evolving, explicit acceptable-use framework detailing exactly what is permitted and what is strictly off-limits.

- Organizations hold full legal accountability for data breaches, copyright infringements, and regulatory infractions caused by generative AI tools. - Mental health and healthcare organizations risk direct HIPAA violations if employees input identifiable progress notes into public models. - The widely publicized legal case involving fabricated precedents highlights the necessity of training staff on algorithmic hallucinations and fact verification.

---

[16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) Critical Questions for Leaders on Workforce Readiness

Transforming organizational culture around AI requires leaders to honestly assess whether their technological deployments are moving faster than their human support systems. Traditional feedback channels like passive suggestion boxes, open office hours, or generic email inboxes fail because workers who fear job loss will rarely volunteer their anxieties in identifiable, public formats.

To bypass middle management bottlenecks that can inadvertently suppress innovation and hide team-level distress, leaders must design proactive engagement initiatives. Options include running dedicated leadership academies that invite bottom-up innovators directly into strategic conversations, as well as conducting randomized, casual check-ins across diverse employee tiers. Equipping teams for the future demands not merely purchasing software licenses, but carving out meaningful, protected practice time and clear strategic goals.

- **Deploying vs. Preparing**: Are you implementing AI tools at a faster pace than you are psychologically and technically equipping your workforce? - **Psychological Safety**: Are feedback loops reliant on passive comment boxes, or are leaders actively seeking out candid concerns regarding job security and ethics? - **Empowered Practice**: Are staff provided protected time, clear prompts, and structured use cases, or are they expected to master complex tools entirely on their own time?

## Notable quotes > [00:04](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=4s) What happens when people are expected to use artificial intelligence at work but have never been taught how to understand or engage with it? — Kari McCoy

> [00:04](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=4s) who is entering the labor force AI-literate and equipped with these tools, and who is being locked out? — Kari McCoy

> [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) sound more like me — Kari McCoy

> [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) putting my head in the sand — Kari McCoy

> [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) No, no, try this way. — Kari McCoy

> [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) Hey, we got a new tool. You're going to use it. It's going to be great. Let's be energized about it, — Kari McCoy

> [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) I knew this wouldn't work. — Kari McCoy

> [13:14](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=794s) have at it. — Kari McCoy

> [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) Are you implementing AI tools at a faster pace than you are psychologically and technically equipping your workforce? — Kari McCoy

> [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) Are feedback loops reliant on passive comment boxes, or are leaders actively seeking out candid concerns regarding job security and ethics? — Kari McCoy

> [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) Are staff provided protected time, clear prompts, and structured use cases, or are they expected to master complex tools entirely on their own time? — Kari McCoy

## Mentioned Kari McCoy, CSU Northridge, Ai4, Generation Alpha, ChatGPT, Apple Siri, HIPAA

## Video description Kari McCoy, Doctoral Candidate/ Career Counselor, CSU Northridge

What happens when people are expected to use artificial intelligence at work but have never been taught how to understand or engage with it? In this talk, I will share findings from a pilot study showing how a brief AI industry panel and hands-on prompt engineering workshop helped continuation high school students move from fear and misinformation toward practical confidence. I will explain how honest conversations about AI, relevant examples, and guided practice can build self-efficacy without minimizing real concerns about privacy, ethics, or job displacement. Attendees will leave with a clear, transferable model for designing AI literacy programs that prioritize psychological readiness as well as technical skills. This talk will help educators and organizational leaders turn AI resistance into informed, responsible action.

[Source](https://www.youtube.com/watch?v=ItMW9xIIpqU)

中文简报

McCoy发现：自上而下的企业AI推行受阻，采用率停滞在20% 若在推行人工智能时未能化解对工作保障的担忧或提供指导性提示词练习，将引发无声的抵触，导致员工采用率止步于20%。 McCoy指出了企业对AI的乐观预期与大众普遍担忧之间日益扩大的鸿沟，并指出由于对失业的恐慌、伦理顾虑以及令人沮丧的早期幻觉问题，缺乏指导的自上而下式推广往往停滞在约20%的采用率。基于针对历来服务不足的延续高中学生开展的试点研究，她阐述了如何通过专家座谈和指导性提示词工程揭开AI的神秘面纱，促使60%的学生走出末日般的恐惧，转向积极且由同伴带动的解决问题模式。将这些见解延伸至职场，她强调组织必须将明确且合规的使用政策（防范HIPAA违规与法律制裁等风险）与结构化、低风险的实践及同伴倡导者相结合。最终，McCoy敦促领导者务必确保人员支持体系与心理建设举措的推进速度与其技术实施同步演进。 AI 素养已成为雇主最看重的头号技能，86% 的雇主预计到 2030 年将借助 AI 转型业务，确保弱势群体获得平等培训是应对职场分化的核心。 (0:04) 职场对 AI 的抵触源自对资源消耗、知识产权侵犯及岗位替代的合理担忧，若缺乏明确制度与心理安全感，员工往往因害怕受罚而隐藏使用或拒绝参与。 (2:25) 针对延续性高中学生的试点研究显示，结合行业专家访谈与针对模拟面试、简历优化的提示词工作坊，成功让 60% 曾抱有科幻式恐惧的学生建立起实用自信与伦理边界。 (7:19) 缺乏针对性实操指导的自上而下推行往往仅能达到约 20% 的采纳率，员工在遇到 AI 幻觉或错误后极易放弃，亟需管理层主动沟通并阐明技术赋能定位。 (10:58) 机构须对第三方 AI 工具引发的数据泄露与合规责任全额买单，建立透明的使用规范并辅以同行互助机制，是防范医疗 HIPAA 违规或法律案例虚构等重大风险的根本保障。 (14:45)

Reading summary

Kari McCoy argues that sustainable workplace AI adoption depends on proactive literacy interventions and psychological safety rather than top-down software mandates. Without structured guidance, clear guardrails, and explicit protections, organizations face silent employee resistance and risk deepening equity divides for historically underserved populations. Workforce Equity Gaps and the Roots of Resistance (0:04) McCoy highlights a stark contrast between tech-conference optimism and workplace realities: while 86% of employers expect to transform their businesses through AI by 2030, and cite AI literacy as their primary incoming skill requirement, underserved workers and lower-socioeconomic students are being systematically left behind. Resistance among workers and students stems from legitimate apprehensions, including data center water and energy consumption, copyright theft, artist displacement, and fears of immediate redundancy. Compounding this resistance is institutional ambiguity, where conflicting policies across classrooms and departments leave workers unsure whether disclosing AI usage will result in disciplinary action or praise for productivity. McCoy observes that unguided top-down mandates—announcing tools without establishing psychological readiness—typically stall at roughly 20% adoption, as hesitant employees encounter early prompt hallucinations and quietly abandon the tools. Demystifying AI Through Hands-On Student Interventions (4:41) To examine how targeted training overcomes avoidance, McCoy conducted a qualitative pilot study with credit-deficient continuation high school students. At the baseline, 60% of students expressed fear that AI was an apocalyptic force destined to take over the world, conflating the entire domain strictly with ChatGPT. The intervention deployed a two-stage model: an industry expert panel that demystified technical limits and ethical issues, followed by a hands-on prompt engineering workshop focused on mock interviews and resume optimization. The workshop demonstrated that guided prompt iteration transformed passive skepticism into active agency. Rather than abandoning tools after early inaccuracies, students learned contextual prompting and began spontaneously coaching their peers. Crucially, the intervention led participants to articulate self-imposed ethical boundaries regarding data ownership, privacy, and academic honesty. Guided Practice, Peer Champions, and Adoption Pathways (13:14) McCoy stresses that moving employees toward fluency requires structured, low-stakes environments rather than simply distributing software licenses. Because unguided staff frequently struggle with where to begin, effective workforce upskilling relies on pairing learners with experienced colleagues, curating short instructional workflows, and leveraging internal peer champions. Organizations can also stimulate organic adoption through structured incentives. McCoy points to departmental showcases and competitions—such as counseling staff competing for prizes by designing practical student outreach solutions—as mechanisms that convert intimidating technologies into collaborative experiments while surfacing high-value workflows. Acceptable-Use Governance and Institutional Liability (14:45) McCoy emphasizes that legal and regulatory accountability rests on the deploying organization rather than third-party software vendors. In regulated sectors like mental health counseling, well-meaning but untrained workers risk severe compliance breaches, such as an intern inputting confidential client progress notes into ChatGPT and inadvertently triggering severe HIPAA violations. Similar hazards occur in professional services, exemplified by an attorney facing financial sanctions after submitting court filings with fabricated precedents generated by model hallucinations. McCoy argues that institutions must establish explicit, evolving acceptable-use frameworks that clearly define prohibited actions, data privacy rules, and fact-verification requirements. McCoy concludes that leaders must honestly evaluate whether technological deployments are outrunning human readiness, cautioning that passive suggestion boxes fail to surface deep anxieties about job displacement. Equipping a workforce requires leaders to actively engage all employee tiers, demystify organizational roadmaps, and provide dedicated, protected practice time.

精炼阅读摘要

Kari McCoy 指出，随着 86% 的雇主预期在 2030 年前通过人工智能重塑业务，AI 素养已跃升为首要求职技能；然而，自上而下的行政摊派极易因员工焦虑与缺乏指导而遭遇抵制，组织必须通过消除技术恐惧、确立制度规范与开展实操练习来构建具备心理安全感的赋能路径。 就业准入鸿沟与普遍存在的道德焦虑 (0:04) 结合在加州社区学院与基础教育阶段的咨询与策略经验，McCoy 指出科技行业内部普遍存在“AI 已被无缝接纳”的认知盲区，而来自较低社会经济背景的弱势群体正面临被排斥在外的系统性风险。根据其引用的数据，86% 的雇主预计到 2030 年将借助 AI 转型业务，AI 素养已成为用人单位招募新员工的核心指标。劳动力市场因而面临根本性的公平拷问：究竟谁能带着完备的技术素养入职，谁又会在门槛外被边缘化。 在讨论提升采纳率前，管理者必须直面一线员工与 Alpha 世代的现实抵触。公众对生成式 AI 的疑虑源于实质性关切：数据中心对水能与电力的高昂消耗、对艺术创作者版权与生计的侵犯、学术剽窃风险以及数据隐私隐患。许多人甚至认为生成式技术在伦理上全然不可靠。若组织未能出台统一的指导规范，导致不同部门或教室各行其是，员工就会因担忧暴露使用会导致失业或问责，进而选择彻底冷漠或隐瞒使用。 破除科幻恐慌：高中的干预试点研究 (4:41) 为了验证从抗拒转向素养的干预路径，McCoy 在面临学分短缺且背景边缘化的继续教育高中学生群体中开展了质性试点研究。干预前，约 60% 的受访学生对 AI 抱有科幻式的末日恐惧，认定该技术必将毁灭世界，并将 AI 狭隘地等同于 ChatGPT，完全未意识到 GPS 导航与语音助手同样基于 AI 架构。该干预分为两个阶段：首先组织行业专家座谈，公开坦陈技术局限、环境成本与伦理陷阱，打破技术神秘感并回应学生对数据抓取与艺术职业替代的真实担忧。 第二阶段为提示词工程工作坊，引导学生围绕模拟求职面试、简历润色及奖学金检索等具体高风险任务展开实操。许多学生曾因初期遭遇模型幻觉而断定技术无用，通过学习添加上下文与迭代提示词，他们掌握了让输出“更像自己”的方法。质性评估显示，学生迅速从回避转为探索实际工作场景，自发为个人数据划定伦理底线，并在熟练后主动指导落后同伴。这种自发的互助现象证实，情境化的引导实操能够有效建立自主掌控感。 自上而下推广的局限与同伴赋能机制 (10:58) McCoy 强调，仅靠领导层高调宣布上线新系统、缺乏指导的自上而下式推行，采纳率通常会停滞在 20% 左右。未受培训的员工往往在尝试输入单条简略指令并得到错误输出后迅速放弃，形成沉默抵制。要逆转这一被动局面，管理者必须主动与各层级员工沟通，了解基层对于被岗位替代的真实恐惧，并明确澄清引入 AI 的实际定位是协助处理繁杂事务、提升团队整体竞争力，而非削减编制。 将技术认知转化为实际信心需要依靠低风险的持续实践。组织应当通过结对互助、短视频培训、收集既有应用案例，以及树立基层同伴标杆来引导员工起步。此外，通过设立内部创新展演或有奖竞赛——例如由部门主管举办优化学生服务的 AI 应用评比，并为专业人员提供专属探索时间——能够将令人望而生畏的技术转化为充满协作感的趣味实验。 机构法律合规责任与领导层自审机制 (14:45) 组织在推行 AI 时必须承担最终的法律责任，第三方软件厂商并不承担违规后果。在心理咨询与医疗等强监管领域，未经训练的人员极易违规将包含敏感个人信息的记录直接输入公共模型，进而引发严重的 HIPAA 隐私违法事件；类似地，法律行业中律师因直接提交包含 ChatGPT 虚构案例的法庭简报而遭受巨额经济处罚的案例，充分印证了培训员工识别算法幻觉与核验信息的紧迫性。管理层必须制定动态更新的合规使用政策，界定操作边界。 McCoy 最后提出，组织文化的转型要求领导者诚实反思技术部署速度是否已经超越了人员心理与技能的准备步伐。传统的意见箱等被动沟通渠道无法打消员工对职业安全的顾虑，管理者需要通过主动设立领导力研修项目或进行跨层级抽样走访，绕开中层信息阻隔。推动劳动力就绪不能止步于采购软件许可，更需要为员工提供受保护的学习时间与清晰的战略指引。 McCoy 总结认为，实现职场向 AI 流畅过渡的核心，在于构建消除失业恐惧的心理安全感、完善机构合规责任并建立同伴支持机制，以防止弱势群体进一步掉队；不过，本报告的实证依据主要源自针对继续教育高中学生的定性试点，其干预模式在跨行业大规模企业生态中的长期适应性与成本效益，仍有待更广泛的实践检验。

Detailed notes — zh

## Background 本视频由 Ai4 频道于 2026 年 8 月 14 日发布在 YouTube 平台上。加州州立大学北岭分校（CSU Northridge）博士候选人兼职业顾问 Kari McCoy 介绍了其开展的一项试点研究，探讨如何通过 AI 行业专家座谈会与实操性提示词工程工作坊，帮助面临辍学风险的进修高中学生克服对 AI 的恐惧与误解，并为教育者和组织管理者提供了一套兼顾心理准备度与技术素养的 AI 素养干预模型。 ## TL;DR McCoy 指出，AI 素养虽已成为关键职场技能，但边缘化群体由于缺乏接触渠道、存在伦理顾虑以及心理安全感缺失，往往对技术产生沉默抵触与恐惧。通过在一所进修高中开展涵盖专家解惑与求职实操（如简历优化、模拟面试）的干预实验，研究成功让 60% 最初认为“AI 毁灭世界”的学生转变为能够建立自我伦理防线的主动应用者。演讲进一步提醒管理者，简单粗暴地强制推行工具会导致采纳率低下，组织必须主动倾听员工担忧、培养内部领跑者并明确法律与伦理红线，确保人员的心理和技能准备能够跟上技术部署的步伐。 ## Atmosphere & style 整场演讲呈现出务实、真诚且充满人文关怀的基调。讲者结合自身在基础教育、高等院校与行业咨询的多维经验，以通俗生动的案例将宏大的技术变革拉回到真实的个体心理与社会公平维度。叙事节奏层层递进，从剖析员工和学生对技术的迷茫与防御心理，到展现动手实践带来的自信赋能与同伴互助，既有对现实伦理风险的冷静审视，又传递出启发人心、注重实操落地的教育热忱。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=4s) 尽管预计到2030年将有86%的雇主通过AI转型业务，但低收入和边缘化背景群体的AI接触与认知严重滞后，加剧了素养鸿沟。 - [02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) 年轻人及员工对AI的抵触主要源于对环境消耗、知识产权剽窃、原创性丧失以及岗位被取代的深层伦理与职业焦虑。 - [02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) 缺乏清晰的使用规范与心理安全感会导致消极抵制，员工往往因害怕被判定岗位无价值而不敢公开使用AI。 - [04:41](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=281s) 进修高中的试点研究表明，许多初学者在遇到一次错误回答后便彻底放弃AI，需要通过系统性引导破除认知误区。 - [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) 通过行业专家座谈与提示词工程工作坊的双重干预，学生从初期的宏观恐惧转变为主动将AI用于模拟面试与简历优化等职业场景。 - [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) 干预使学生获得了伦理自我赋权（Ethical Agency），不仅消除了盲目恐惧，还学会了保护个人数据隐私并建立合理的使用边界。 - [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) 自上而下强制推行工具往往只能带来约20%的低采纳率，管理者必须主动倾听各层级顾虑，明确AI是赋能而非替代。 - [13:14](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=794s) 组织需要提供结构化的引导式起步，并通过竞赛奖励机制培养内部领军者（Champions）以带动同伴互助学习。 - [14:45](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=885s) 企业必须将法律合规与伦理防线融入培训，严防员工因大模型幻觉导致法律风险，或因输入敏感数据违反HIPAA等隐私法规。 - [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) 领导者应反思技术部署速度与员工心理准备度之间的脱节，通过突破层级的沟通机制与充足的实操时间构建长效准备度。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=4s) 演讲背景与AI素养鸿沟 演讲者 Kari McCoy 首先分享了自己的多元背景。她目前在两所社区学院担任退伍军人与职业顾问，拥有长达 10 年的 K-12 阶段教师与高中咨询顾问经历，同时还担任一家非营利机构的主席，并兼任 AI 战略与政策顾问。正是这种跨越基础教育、高等教育与行业咨询的视角，促使她关注一个核心议题：企业和教育机构究竟该如何有效提高人员对 AI 技术的接纳度（Adoption）？ McCoy 指出，业内人士常常生活在一个信息茧房（bubble）中，误以为每个人都在热情拥抱并频繁使用 AI。然而现实远非如此，尽管业界普遍认为到 2030 年将有 86% 的雇主通过 AI 转型其业务，且 AI 素养已被列为头号紧缺技能，但社会经济地位处于劣势、历史上常被边缘化的学生群体却极少能获得同等的接触机会与认知渠道。如果这部分群体在进入职场前缺乏 AI 素养，他们与同龄人之间的竞争力差距将会进一步被拉大。 - **86% 雇主预期**：预计在 2030 年前利用 AI 实现业务与运营模式的全面转型。 - **核心痛点**：AI 素养成为最关键的就业技能，但低收入与边缘化背景群体的 AI 接触率与认知度严重滞后。 --- ### [02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) AI抵触情绪与职场领导力的挑战 在推行 AI 技术之前，必须先理解决策对象内心的恐惧与抵触从何而来。McCoy 强调，包括 Alpha 世代在内的许多年轻群体对 AI 持有极大的怀疑与伦理顾虑。他们普遍担忧 AI 正在消耗大量水资源、破坏生态环境、剽窃艺术家的创意成果，甚至认为任何 AI 使用本质上都是不道德的抄袭。如果在没有解决这些疑虑的情况下强推技术，只会招致强烈的被动抵抗。 除了伦理担忧，缺乏明确规则与“心理安全感”（psychological safety）也是阻碍 AI 普及的关键因素。许多学生和职场新人常常感到无所适从：“在一个课堂上老师允许使用，在另一个课堂上却被明令禁止；我不敢告诉老板我在用 AI，因为担心老板会觉得我的岗位不再有价值，但又怕不用 AI 会显得缺乏工作效率。” McCoy 认为，管理者必须在制度层面明确 AI 使用的边界与预期，让员工确信使用 AI 是为了提升效能而非取代其工作。 ```mermaid flowchart TD accTitle: AI Adoption Barriers Flowchart accDescription: Flowchart illustrating the primary concerns and leadership ambiguities hindering AI adoption. subgraph Concerns ["员工与学生的底层顾虑"] C1["伦理与原创争议<br/>(抄袭/窃取创意)"] C2["环境与资源消耗<br/>(数据中心耗水/能耗)"] C3["职业替代恐惧<br/>(失去工作岗位)"] end subgraph LeadershipGaps ["管理与政策缺失"] L1["规则模糊不清<br/>(允许与禁止边界不明)"] L2["心理安全感匮乏<br/>(不敢向主管坦白使用)"] end Concerns --> Resistance["抵触与回避行为<br/>(拒用/装作看不见)"] LeadershipGaps --> Resistance ``` --- ### [04:41](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=281s) 针对进修高中学生的AI素养干预实验 为了探索如何打破这种心理壁垒，McCoy 在一所“进修高中”（Continuation High School，即针对学分缺失、面临辍学风险或传统学校体系未能充分支持的边缘化群体设立的高中）开展了一项质性干预研究。在干预初期，多数学生将 AI 视为一种遥远且具有威胁性的“科幻产物”（sci-fi），并普遍存在认知误区，简单地将 AI 与 ChatGPT 画上等号，完全不知道日常生活中的 GPS 导航、Siri 语音助手甚至部分日常消费品中都早已嵌入了 AI 技术。 研究团队设计了两个关键干预阶段： 1. **AI 行业专家座谈会（Panel Discussion）**：邀请 AI 从业人员与学生面对面，坦诚讨论 AI 是什么、不是什么、其局限性与伦理考量，以及如何运用 AI 增强就业竞争力。 2. **实操性提示词工程工作坊（Prompt Engineering Workshop）**：指导学生利用 AI 工具进行模拟面试演练、简历优化以及寻找入门岗位和奖学金机会，将抽象技术转化为可感知的生存技能。 --- ### [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) 关键研究发现：从恐惧回避到负责任的主动应用 研究表明，许多学生最初放弃 AI 是因为单次糟糕体验——“我之前在 ChatGPT 里输了一个问题，它给了我一个错误答案，于是我就彻底不用了，觉得它根本没有大家吹得那么神奇”。然而，在经过提示词技巧与实操演练后，学生的态度发生了戏剧性转变： ```mermaid flowchart LR accTitle: Transition from Fear to Fluency accDescription: Stages of student mindset shift from fear and avoidance to responsible empowerment. A["初始阶段: 恐惧与回避<br/>(60% 认为 AI 毁灭世界)"] --> B["阶段一: 专家交流破除迷思<br/>(认知局限与现实用途)"] B --> C["阶段二: 提示词实操演练<br/>(简历优化/模拟面试)"] C --> D["最终状态: 主动探索与伦理赋权<br/>(互助分享/自设使用边界)"] ``` 干预实验带来了三个维度的显著转变： - **从宏观恐惧到落地应用**：干预前，约 60% 的学生认为“AI 即将毁灭/统治世界”（AI was going to take over the world）；干预后，他们学会了通过多轮迭代提示让 AI 输出符合自己语言风格的内容，开始认识到工具的实际价值。 - **从消极回避到主动探索**：学生从把头埋在沙子里的抗拒态度，转变为积极发问并尝试将 AI 用于职业规划，例如利用 AI 检索入门级岗位、根据个人简历匹配工作或申请大学奖学金。 - **从伦理愤怒到自我赋权（Ethical Agency）**：学生不再仅仅因为数据隐私和版权问题感到愤怒和排斥，而是学会了如何保护个人隐私并建立属于自己的伦理使用边界（"This is how I'll use it to protect myself"）。 在工作坊现场甚至出现了生动的同伴教学现象——掌握更快的学生主动教身边的同学：“不不，你试试这种提问方式！”这种自信心最终转化为对技术的负责任应用。学生们在活动结束时能够清晰规划自己将如何使用 AI，并主动为其设定合理的行动边界。 ### [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) 正视恐惧与抵触：从被动排斥到建立信任 在推行新技术时，如果领导层未能主动正视员工内心的恐惧，这种变革往往会遭遇极大的阻力。正如前面讲者 Belinda 所提到的，如果管理层只是简单地告诫员工“AI 时代已经到来，保住工作的唯一方法就是融入其中”，或者只是扔下一个新软件并轻描淡写地说：“嘿，我们上线了一个新工具，大家快用起来，这太棒了！”，实际的采纳率往往只有区区 20% 左右。除了少数乐于尝试的早期采纳者，绝大多数人都会选择“默默抵制”。在缺乏系统指导的情况下，员工可能会试着用 ChatGPT 提一个问题，一旦得到不满意的糟糕回答，他们就会立刻放弃并验证自己的偏见：“看吧，我就知道这玩意儿根本不管用。” 相反，领导者必须主动“走到员工所在的位置”，倾听并接纳他们的担忧。管理者应当主动询问各个层级的利益相关者——无论是初入职场的实习生，还是资深的中层工程师：“你们对 AI 的真实感受是什么？”为他们创造坦诚表达顾虑的空间。只有摸清员工的真实想法，才能对症下药。比如，当发现员工误以为“引入新工具就等于明天要被裁员”时，管理层就可以明确澄清工具的实际定位和应用范围，打破信息迷雾。当员工明白引入 AI 是为了提升生产力和团队竞争力，而非取代他们时，恐惧感就会被彼此支持的“战友式情谊”（camaraderie）所取代。 ```mermaid flowchart TD accTitle: AI 推行路径对比 subgraph FailurePath["自上而下施压（失败路径）"] A["直接推行工具\n'大家都得用'"] --> B["缺乏引导与指导\n试用受挫"] B --> C["沉默抵制与恐惧加剧\n采纳率仅约 20%"] end subgraph SuccessPath["以人为本引导（成功路径）"] D["主动倾听与正视顾虑\n询问实习生/工程师感受"] --> E["消除误解明确范围\n强调赋能而非替代"] E --> F["引导式练习与同伴赋能\n形成创新文化与高采纳度"] end ``` - **低效推行的典型采纳率**：在缺乏心理建设与指导的强制推行下，新技术采纳率通常低至 20% 左右。 - **沉默抵制的心理机制**：缺乏引导的初次负面体验（如得到一次错误的生成结果）会迅速演变成放弃与抗拒借口。 - **有效沟通的核心要点**：通过分层访谈明确推行 AI 的真正目标——提高人效与组织竞争力，而非短期人员替换。 --- ### [13:14](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=794s) 引导式实践与同伴激励：寻找内部领军者 光有沟通还不够，缺乏对工具的理解和信心是阻碍采纳的另一大核心痛点。如果只是把工具扔给员工说一句“随便试”，大多数人根本不知道从何下手。组织必须提供清晰的“引导式起步”（guided starts），例如提供入门视频、开展专项调研以发掘哪些员工已经在自发使用，并总结出最佳实践。更为关键的是，组织要善于在内部发掘并培养“领军者”（champions）与“影响力人物”（influencers），让他们带动身边的同事。 为了激发全员的创新活力，管理层可以通过竞赛和奖励机制来调动积极性。例如，在咨询顾问团队中，主管可以设立一项挑战赛：“谁能找到最适合辅助我们服务学生的 AI 工具并进行展示，谁就能获得丰厚奖励。”这种机制不仅能激励创新，还能让涌现出来的领跑者成为同伴导师。无论是软件工程团队还是日常业务部门，都需要为员工提供充足的“玩转工具的时间”（time to play），并借助同行之间的经验分享，让大家在安全、轻松的氛围中获得实操技能。 - **破除起步障碍**：提供结构化的引导素材与操作范例，避免员工面对空白输入框无所适从。 - **内部冠军网络**：利用同伴影响力和榜样力量，让早期探索者分享实操技巧以带动周围同事。 - **正向激励机制**：通过举办创新应用大赛、设立专项奖项等方式，将工具探索转化为被认可的组织行为。 --- ### [14:45](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=885s) 将伦理防线与法律责任融入全员培训 在推动 AI 应用的同时，必须将伦理与合规要求深深植入培训之中。许多企业和机构目前正面临与学校类似的困境——缺乏统一明确的 AI 使用规范。比如一家心理健康诊所非营利机构就曾求助咨询，因为他们完全不知道该如何制定机构内部的 AI 政策。正如前面讲者 Georgia 所强调的问责问题：“出现问题时，究竟谁该承担责任？”答案毫无疑问是企业自身，法律责任并不会因为使用了 AI 工具而免除。 如果缺乏明确的伦理边界培训，员工极易踩中严重的合规红线。例如，心理健康机构的实习生如果擅自将客户的隐私档案输入 ChatGPT 来生成病程记录（progress notes），就会直接构成严重的 HIPAA（健康保险可携性与责任法案）违规。再比如法律界广为人知的真实案例：一名律师使用 ChatGPT 检索判例并直接提交给法庭，结果因引用了大量 AI 虚构、无法核实的假判例而面临巨额罚款。普通员工往往意识不到 AI 存在数据隐私泄露和准确度缺陷等局限性，因此领导层必须建立清晰、与时俱进的伦理规范，明确告知员工“什么场景可以用、什么场景坚决不能用”。 - **非营利与医疗机构的隐私风险**：严防员工将涉及 HIPAA 等受保护的敏感数据输入未经授权的公共大模型。 - **虚构判例的前车之鉴**：警惕大模型“幻觉”导致的真实法律与财务损失，严禁不加核实地采纳输出结果。 - **动态更新的政策机制**：AI 规范必须随着技术发展与业务场景变化保持敏捷迭代，同时守住法律底线。 --- ### [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) 领导者的核心反思与长效机制 在演讲的尾声，针对希望在组织内推进 AI 准备度（AI readiness）的管理者，主讲人提出了几个直击本质的核心反思问题： > “你们组织部署 AI 的速度，是否远远超过了为员工做好心理和技能准备的速度？” > “你们是否真正为员工坦诚探讨伦理与职业担忧创造了安全空间？” 仅仅设置一个匿名意见箱、发送一封全员邮件或安排常规答疑时间（office hours）是远远不够的。那些深陷“工作被取代”恐慌中的员工，绝不可能通过这些常规渠道主动敞开心扉。领导者需要采取更有创意的沟通方式：例如建立“领导力学院”（Leadership Academy），让一线的创新者能够绕过可能扼杀创新的中层管理障碍，直接与高层对话；或者在全员中进行随机抽样沟通（random sample check-ins），真正摸清基层员工的真实顾虑与技能水平。最重要的是，组织是否为员工提供了有意义的上手实操机会，给予他们充足的时间去摸索，而不是简单地派发一份新软件清单便撒手不管。 - **部署速度与准备速度的平衡**：警惕“技术推进快、人员准备慢”导致的脱节与内耗。 - **打破层级阻碍**：通过创新机制让一线员工的声音和金点子能够直达管理高层。 - **持续对话与赋能**：从单向通知转向有意义的实操赋能，帮助组织在 AI 浪潮中建立真正的韧性。 ## Notable quotes > [02:25](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=145s) 在一个课堂上老师允许使用，在另一个课堂上却被明令禁止；我不敢告诉老板我在用 AI，因为担心老板会觉得我的岗位不再有价值，但又怕不用 AI 会显得缺乏工作效率。 — Kari McCoy > [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) 我之前在 ChatGPT 里输了一个问题，它给了我一个错误答案，于是我就彻底不用了，觉得它根本没有大家吹得那么神奇 — Kari McCoy > [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) AI was going to take over the world — Kari McCoy > [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) This is how I'll use it to protect myself — Kari McCoy > [07:19](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=439s) 不不，你试试这种提问方式！ — 进修高中学生 > [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) AI 时代已经到来，保住工作的唯一方法就是融入其中 — Belinda > [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) 嘿，我们上线了一个新工具，大家快用起来，这太棒了！ — Kari McCoy > [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) 看吧，我就知道这玩意儿根本不管用。 — Kari McCoy > [10:58](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=658s) 你们对 AI 的真实感受是什么？ — Kari McCoy > [13:14](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=794s) 谁能找到最适合辅助我们服务学生的 AI 工具并进行展示，谁就能获得丰厚奖励。 — Kari McCoy > [14:45](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=885s) 出现问题时，究竟谁该承担责任？ — Georgia > [14:45](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=885s) 什么场景可以用、什么场景坚决不能用 — Kari McCoy > [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) 你们组织部署 AI 的速度，是否远远超过了为员工做好心理和技能准备的速度？ — Kari McCoy > [16:22](https://www.youtube.com/watch?v=ItMW9xIIpqU&t=982s) 你们是否真正为员工坦诚探讨伦理与职业担忧创造了安全空间？ — Kari McCoy ## Mentioned Kari McCoy, Belinda, Georgia, CSU Northridge, ChatGPT, Siri, GPS, HIPAA ## Video description Kari McCoy, Doctoral Candidate/ Career Counselor, CSU Northridge What happens when people are expected to use artificial intelligence at work but have never been taught how to understand or engage with it? In this talk, I will share findings from a pilot study showing how a brief AI industry panel and hands-on prompt engineering workshop helped continuation high school students move from fear and misinformation toward practical confidence. I will explain how honest conversations about AI, relevant examples, and guided practice can build self-efficacy without minimizing real concerns about privacy, ethics, or job displacement. Attendees will leave with a clear, transferable model for designing AI literacy programs that prioritize psychological readiness as well as technical skills. This talk will help educators and organizational leaders turn AI resistance into informed, responsible action. [Source](https://www.youtube.com/watch?v=ItMW9xIIpqU)

Source quotes

[2:25] Kari McCoy: 在一个课堂上老师允许使用，在另一个课堂上却被明令禁止；我不敢告诉老板我在用 AI，因为担心老板会觉得我的岗位不再有价值，但又怕不用 AI 会显得缺乏工作效率。 [7:19] Kari McCoy: 我之前在 ChatGPT 里输了一个问题，它给了我一个错误答案，于是我就彻底不用了，觉得它根本没有大家吹得那么神奇 [7:19] Kari McCoy: AI was going to take over the world [7:19] Kari McCoy: This is how I'll use it to protect myself [7:19] 进修高中学生: 不不，你试试这种提问方式！ [10:58] Belinda: AI 时代已经到来，保住工作的唯一方法就是融入其中 [10:58] Kari McCoy: 嘿，我们上线了一个新工具，大家快用起来，这太棒了！ [10:58] Kari McCoy: 看吧，我就知道这玩意儿根本不管用。 [10:58] Kari McCoy: 你们对 AI 的真实感受是什么？ [13:14] Kari McCoy: 谁能找到最适合辅助我们服务学生的 AI 工具并进行展示，谁就能获得丰厚奖励。 [14:45] Georgia: 出现问题时，究竟谁该承担责任？ [14:45] Kari McCoy: 什么场景可以用、什么场景坚决不能用 [16:22] Kari McCoy: 你们组织部署 AI 的速度，是否远远超过了为员工做好心理和技能准备的速度？ [16:22] Kari McCoy: 你们是否真正为员工坦诚探讨伦理与职业担忧创造了安全空间？

Source quotes

[0:04] Kari McCoy: What happens when people are expected to use artificial intelligence at work but have never been taught how to understand or engage with it? [0:04] Kari McCoy: who is entering the labor force AI-literate and equipped with these tools, and who is being locked out? [7:19] Kari McCoy: sound more like me [7:19] Kari McCoy: putting my head in the sand [7:19] Kari McCoy: No, no, try this way. [10:58] Kari McCoy: Hey, we got a new tool. You're going to use it. It's going to be great. Let's be energized about it, [10:58] Kari McCoy: I knew this wouldn't work. [13:14] Kari McCoy: have at it. [16:22] Kari McCoy: Are you implementing AI tools at a faster pace than you are psychologically and technically equipping your workforce? [16:22] Kari McCoy: Are feedback loops reliant on passive comment boxes, or are leaders actively seeking out candid concerns regarding job security and ethics? [16:22] Kari McCoy: Are staff provided protected time, clear prompts, and structured use cases, or are they expected to master complex tools entirely on their own time?

Original transcript

[0:04] Thankfully, the presentations that we had right  before us really merge well with what I'm going [0:10] to talk to you all about today. And I hope I  don't cross over too much of what has already [0:15] been said. I'll try and be pretty concise and go  different directions that are most useful for you. [0:20] But the focus I'm going to go over is mainly  adoption increasement increasing. So how do [0:26] you increase adoption of your employees for AI  technology? But the focus of how we got there [0:32] is going to be kind of interesting. Okay, so to  give you a little bit of background on who I am, [0:36] my name is Carrie McCoy. I currently am a veterans  and career counselor at two community colleges. [0:42] I also have 10 years of K-12 experience working  as a teacher and a high school counselor. [0:48] And then I have some non-profit experience.  And so I've created initiatives and I've run [0:53] in the non-profit sector. I'm president of a  non-profit. I also work as an AI strategy and [0:59] policy consultant. So that just gives you kind of  a background on where my lens is as I go. I'm I'm [1:04] just going to pretend to pass in things. Um So you  all know those who are here obviously know that AI [1:13] is not coming, it's already here, right? It's very  present in our life and the dependency is growing [1:18] every day. But what you may not know is that not  everyone is using it. You know, there's kind of a [1:23] misnomer when we're in this bubble that everyone's  about it and everyone understands that this is [1:29] here and going to stay and that they're going to  have to embrace it. But that's not commonplace [1:34] and what I was seeing and what I  was interested in understanding was [1:38] you know, are all students using it? Because  what if some students that are being left behind [1:45] historically in lower socioeconomic statuses are  not being given the same access or understanding [1:52] and then what is their adoption rate? And  how could would be um, utilized understand [1:57] that for employment? So, we know that 86%  of employers are expecting to transform [2:05] um, their businesses by with through AI by  2030. And that we know that is the number [2:11] one skill that people are looking for, that  AI literacy is so important. But who all is [2:17] coming into the workforce being AI literate?  Who understands those tools and who doesn't? [2:25] The real challenge I don't know if that was me  or you this time. The real AI challenge is who [2:31] who is able to access this information, who's  ready for it, who's scared of it, who's resisting [2:36] it? And we should we should understand that there  is a lot of resistance coming from Gen Alpha and [2:42] from other people that feel like AI is taking our  water, is taking our resources, is not ethical, [2:50] that no usage is ethical, that it is plagiarism,  that is stealing creativity. There's a lot of [2:56] controversial feelings about this. And I think  that before you ask the question of how do I [3:01] get all of my people to adopt this technology,  you need to understand where the apprehension is [3:06] coming from because this is something that people  are worried will change our world and it could it [3:11] could. But also, will they be left behind by not  utilizing it to any extent and understanding it. [3:18] So, my concern from this lens was if some students  are using it and some students are gaining [3:24] gainful employment through the use of these  AI tools, but others are being left behind, [3:29] how do we give them a competitive edge? Um, so  the questions that we want to understand when [3:35] getting people to adopt to AI, it's building their  confidence, it's limiting the ethical usage of it, [3:42] being clear and direct about what usage, and  I'll give you a framework on as an employer [3:48] what to expect to get the best AI adoption rate.  Um, but understanding that practical knowledge [3:55] and as Belinda said, you know, understanding  the psychological safety, that they need to know [3:58] that their jobs aren't going to be replaced  if they start utilizing those tools, right? [4:04] And then what kind of expectation is there from  leadership on what amount to use the tool. You [4:11] know, a lot of what I hear students ask is, "Well,  where's the AI policy? Like, what am I supposed [4:17] to be using it for? Because in one class, the  teacher is allowing me to use it and in another [4:21] I'm not allowed to use it. I don't know if I can  tell my boss that I'm using it or not." You know, [4:26] because maybe my boss will think that my job  should be gone because I am using it, but maybe my [4:31] boss will think that I'm a better worker because  I'm using it and it's making me more productive. [4:36] So, you, from a leadership standpoint, being  clear about that direction is really important. [4:41] So, for this study, what I did was a a qualitative  study where I took students from a continuation [4:47] high school and I gave them interventions to  teach them about AI and to kind of cross over [4:53] what AI literacy should look like. So, we had AI  professionals come in to do a panel discussion [4:59] on what artificial intelligence is and  what it's not, what the limitations are, [5:04] what the ethical concerns are, where it's going,  and how to be um using it to better their chances [5:10] in the workforce. And then they were able to ask  questions. of questions were interesting because [5:15] many students didn't understand what artificial  intelligence is. Right now, students believe, [5:20] not all students, but a good majority of  students feel that AI is synonymous with ChatGPT. [5:26] And that's the extent of it. They don't know that  GPS and uh Siri are uses of AI, right? Or that our [5:33] nail polish might have AI um applicable tools. So,  understanding that was really important before we [5:41] got into the second intervention, which was a  prompt engineering workshop where we gave them [5:46] that skill for their jobs, how to do mock  interviews, how to better their resume, [5:52] how to apply practical usage of AI tools  to their life. And then we interviewed them [5:57] afterwards to understand where did this change  for them their fear or their feelings about AI? [6:04] And what we saw in the beginning was that  there was this generalized fear of what [6:08] artificial intelligence is. There was this kind of  distance from themselves. This AI is sci-fi. AI is [6:15] And I should contextualize that this was last  spring with a continuation high school. So, [6:20] a very specific population if you're wondering  like, what? I think all students are using it, [6:26] but this was looking at students that are  deficit in credits, that were falling behind, [6:31] that were taken out of traditional schools, that  are coming from backgrounds that are typically [6:36] underserved, okay? So, back to what we saw, um  there was this distance where they felt like AI is [6:43] science fiction. It's going to take  over the world. I'm worried about it, [6:46] but I don't understand it, so I'm just going to  stay away from it and put my head in the sand. [6:50] And that was a lot of the sentiment. There was  concerns about creativity. A lot of questions [6:55] that came up was, how are artists going to survive  if all of their work gets stolen? Where is the [7:00] privacy for that? Where is the safety for that?  And what kind of environmental concerns are there? [7:06] Because I hear that it's taking up all the water,  that it's polluting water. So, a lot of concerns [7:12] about what they're hearing in the world, but not  what they're getting invested, not active agency. [7:19] So, as I mentioned, the first thing we did  was we brought panelists to talk to them. [7:24] And then we went to um building of mastery,  where we gave them prompt engineering, [7:29] and we gave them um one-on-one experiences with  how to use these tools and prompt better. So, [7:36] what some of them shared was, initially I had put  into ChatGPT a question, and they gave me a wrong [7:41] answer, so I quit. I was done. I didn't understand  how this was so great, why people talk about it [7:46] as being this magical thing. It didn't work for  me. But then when we did this prompt engineering, [7:51] we saw a different response. So, one thing that  if you're an employer, what you can consider [7:56] is instead of an AI panel talking about like  what artificial is artificial intelligence is, [8:02] you could do that, but you could also bring in  relevant to their context as a job, what their job [8:09] is and how artificial intelligence would look for  their job. So, the first finding was before in the [8:14] before the intervention, we saw that AI was going  to take over the world was kind of the sentiment [8:19] from 60% of the the students. But then afterwards,  they saw how they could use it and they understood [8:26] how other people could use it and they could  understand why it was this magical thing because [8:30] they started seeing, well, if I could use it  for this, then I could apply it over here. [8:34] I didn't realize it could do that. I didn't  realize I could ask it to sound more like me, [8:39] that I could be reiterative. [clears throat] And  then the second the second finding that we saw was [8:45] they went from being fearful and avoidant of the  information and what artificial and tools could do [8:51] to becoming actively engaged with it. So, then  they had more questions. Well, could I use it [8:55] for this or could I use it tomorrow? If I was  going to get a job, could I locate you know, [9:02] entry-level positions? Could I put my resume in  and have it find jobs like mine? Could I find [9:07] scholarships with it? So, they were more actively  engaged with understanding how to use the tools. [9:14] And then the third time finding that we had was  that there was an empowerment with the ethical [9:20] awareness. So, initially when they were scared  that it was going to take every license, they were [9:24] angry, angry about the idea that it would take all  of the artistry out there and that there was data [9:31] privacy concerns and they didn't understand,  why would you use that if it could take all [9:34] of your information, to then understanding what  the limitations were. This is how I'll use it to [9:40] protect myself. This is how I'll use it, but  I will have ethical boundaries for myself of [9:46] what's too much, and I won't rely on other people  to state that. So, again, that active agency, [9:52] but then also feel feeling confident with  understanding where the limitations existed [9:58] and where their boundaries were. So, overall,  what you kind of see is this different stages [10:04] from when I I don't know anything about  it and I don't know how to get my answers [10:09] uh my questions answered. There's this fear and  avoidancy to then a little bit of understanding, [10:15] a little bit more questioning, a little bit more  agency to then when we were in the practice, [10:20] we saw a lot more interest in exploring it and  questioning what else it can do, to then having [10:27] that confidence and that willingness to teach  another. So, during the prompt engineering, [10:32] when somebody got it quicker, they were going to  their neighbor and saying, "No, no, try this way." [10:36] You know, and so that confidence that we saw led  to ideally over responsible usage. So, now they [10:42] come at the end telling us, "Well, I'm going to  use it for this, and I'm going to use it for that, [10:46] and I didn't know I could do that with it." So,  now they're setting boundaries for themselves. [10:51] So, how does this relate to employment?  Well, what I think that you can utilize is [10:58] when we name the fear, and this is a lot to what  Belinda was saying, if you don't acknowledge [11:04] and you say, "Well, I know AI is coming and  that the only way to save their job is that, [11:08] you know, embed themselves in it and to get to  get to work with evolving themselves with it," [11:14] then you have this resistance. You might say,  "Hey, we got a new tool. You're going to use it. [11:20] It's going to be great. Let's be energized about  it." And you send them on their way. That level [11:26] of adoption is like a 20%, right? You're going to  not have You're going to have some early adopters [11:30] that want to show how they can make things better,  but the majority of people will silently resist. [11:36] Maybe they try it, but because you haven't given  them that guidance, they're now going to touch it, [11:42] it's not going to work for them, they're going  to put that question into ChatGPT as an example, [11:46] and get a bad response, and then quit. They say,  "That's up. It's I'm not interested in it. I knew [11:50] this wouldn't work." And they're going to show  their apprehension, they're going to show their [11:53] fear, and their resistance. But, if you instead  meet them where they are, and acknowledge their [11:59] concern, ask them how they're feeling about  it. Go to your stakeholders and ask them, [12:04] "Hey, you're my, you know, intern. How do you  feel about it? You're my, you know, mid-level [12:10] um engineer. How do you feel about it?" And give  them space to talk about their concerns about it, [12:15] because once you gather that intel, and you know  what their concerns are, then you can address [12:20] and validate those concerns. "Oh, you thought it  if if this new tool is coming, it means you're [12:25] getting replaced tomorrow." No, no, actually,  here's the scope of what we want to do with it. [12:31] And that's where you're kind of demystifying what  the plans are. So, if you're clearing clearing out [12:36] and clarifying for them where you want it to  go, what you want them to be capable of, that [12:41] your hope is that they become more productive,  and that together you become more competitive, [12:47] then there's less apprehension, right? They  feel like you have this camaraderie. And I would [12:52] urge you to consider, if any of you have had it  leadership experience, what how do you get buy-in, [12:58] you know, from people who kind of has stake in the  ground when there's changes that come? It's same [13:03] same idea, right? That you want to kind of take  on leaders and champions and influencers, and you [13:09] want to get them to buy in, but you need to first  address their concerns. Because there's a whole [13:14] lot of lack of understanding, and then there's  this lack of confidence. So, then you need to give [13:19] time to practice. Just as Belinda went over all of  these workshops and trainings and voluntary time, [13:26] but if you give a person a tool and say,  "Have at." they don't know where to start. So, [13:31] you need to get them guided on where to start. And  what you can do with that is you can find videos, [13:36] you can find champions, you can survey your  people and ask them like who's using this, [13:41] what best practice do they have, and have them  lead things, incentivize them to come up with, [13:47] you know, ways to use the tool. Like for example,  as a veteran counselor, if my counseling dean said [13:53] to me or said to all the counselors, "Hey, there's  going to be a competition and whoever can find [13:59] an AI tool that best helps us reach students, you  know, showcase we're going to have a showcase and [14:05] there's going to be a prize." Or something like  that, right? Where you incentivize innovation, [14:10] but also finding those champions so that they  can be helping and prioritizing practice, [14:15] getting guided practice. So, if you have  a software engineer and you're trying to [14:20] implement a new tool or create tools with that  software engineer, you need to be finding them [14:27] counterparts that can give them guiding  starts. And then time to play with the tools. [14:33] And then using those peers. So, as I mentioned,  finding those champions and influencers and [14:38] making sure that they're sharing their knowledge  with others and they're trying to gather buy-in. [14:45] And then building the ethics into training. So,  it is imperative, like I mentioned with students [14:50] feeling like there's not consistent policy  across the board, employment is the same way. [14:55] I work with a nonprofit who asked me to look at  their policy because they are mental health clinic [15:01] and they don't know what their AI policy  should be. But those ethics are really [15:07] imperative because as Georgia, earlier speaker,  mentioned, you know, who's accountable, right? [15:12] Well, ultimately the company is. And so, you  need to have a clarified policy that's evolving [15:17] and changing as needed to meet the moment as  it changes. But you're still responsible for [15:23] the same legal responsibilities. So, if you have  HIPAA violations, if you have an intern putting in [15:29] a client's information into ChatGPT and say,  "Give me progress notes for it." That's going [15:33] to be a problem. So, you need to clarify what  are the ethical concerns of your organization. [15:39] If you're a lawyer, for example, I'm sure many of  you have heard this case study, but a lawyer used [15:45] ChatGPT to gather research on precedent, and he  brought that into a court. And now has been fined [15:51] a lot of money because half of that precedent was  illegitimate and wasn't verifiable. And so, these [15:57] are things that if you don't train your people on  what AI's limitations are, they will step into. [16:04] They don't understand necessarily that there are  data privacy concerns, that there are limitations [16:09] with its accuracy. So, it's your responsibility  in leadership to make sure those people understand [16:14] that. So, make sure you have that ethical training  of how you want to use, how you not want to use. [16:22] Um and now we have So, AI readiness in the  workforce, the last thing I'll say to you really [16:27] is some questions that I have for you to think  about if you're a leader in your organization, [16:31] or you would like to be, is are you deploying AI  faster than you're preparing the people for it? [16:39] And then, how are you creating spaces for honest  honest concerns about ethics? So, if you're [16:45] saying, "Well, we have a comment box, or they can  email me. I have office hours that are available." [16:50] That's not going to be enough for those people  who are scared of their jobs being displaced [16:55] for them to come openly to you and share their  concerns. So, I challenge you to think of [16:59] other creative ways. You know, I was having a  conversation with somebody earlier at my poster [17:04] about how if you could do a leadership  academy to invite innovators to come to you [17:11] instead of middle management, you know, because  sometimes that stops innovation. Or coming to your [17:16] people and just doing check-ins with all kinds  of different people and doing a random sample [17:21] of your employees to make sure that you understand  where they are. And then are employees being given [17:26] meaningful opportunities to practice the tools?  Are you just saying here's the new software [17:31] or you giving them time and ideas on how  to use those tools? And that's it for me. [17:37] So you can follow me on LinkedIn and check in with  me. Feel free to I'll be here after if you have [17:43] any questions for me and hopefully that gives you  something to go with go home with and and utilize.